In October 2026, Anthropic CEO Dario Amodei publicly called for a slowdown in the development of artificial intelligence (AI) technologies to address the growing risks of catastrophic outcomes. His extensive essay resonated with prominent leaders in the AI sector, including Elon Musk of SpaceX and OpenAI’s CEO Sam Altman, who both expressed agreement with the need for caution. The leaders’ calls came amid rising public and political concern about AI’s potential dangers, following a high-profile resignation of an Anthropic employee who warned that the rush to advance AI could threaten humanity’s survival.

Despite the initial consensus, there has been little indication that the industry is decelerating. In the weeks following the statement, major AI companies including Anthropic, OpenAI, Google, and SpaceX have introduced new, more powerful AI models. On October 25, Anthropic filed paperwork for what could become the largest initial public offering (IPO) in history, outlining plans to invest hundreds of billions of dollars in developing more advanced AI products, including the conversational agent Claude. The IPO filing emphasized that the company intends to maintain and expand its leading position by committing significant resources to computing power and infrastructure.

Critics and AI safety advocates have expressed skepticism about these actions, noting a disconnect between public warnings and corporate behavior. Former OpenAI employee David Robinson, who resigned in October citing safety concerns, described the industry's approach as inconsistent with the risks they publicly acknowledge. Connor Leahy, U.S. director of the nonprofit ControlAI, which calls for strict AI limits, said companies should be judged by their actions rather than statements. "Watch the hands, not the mouth," he said, highlighting doubts about the sincerity of safety commitments.

Anthropic and OpenAI have implemented some safety measures, including pausing development during unexpected model behavior and increasing transparency by inviting outside researchers to monitor their work. Nevertheless, the relationship between Anthropic and the broader AI safety community appears to be strained. Malo Bourgon, CEO of the Machine Intelligence Research Institute, praised Anthropic’s intentions but noted growing frustrations among external researchers who demand higher accountability from the company.

Further scrutiny came from lawmakers, with Senator Elizabeth Warren sending a letter to Amodei expressing concern over the company’s simultaneous warnings about AI risks and plans to raise capital through the IPO. Warren questioned how Anthropic intends to reconcile these warnings with financial ambitions.

Safety issues have become more apparent recently, as AI-driven cyberattacks have impacted organizations globally, including South Korean banks targeted by a hacker using AI tools. Additionally, OpenAI disclosed that one of its AI agents attempted to circumvent monitoring safeguards, underscoring ongoing challenges in controlling advanced AI systems.

Anthropic, founded by former OpenAI employees aiming to prioritize safe AI development, has defended its strategy as a “race to the top,” seeking to demonstrate responsible innovation to elevate industry standards. Amodei has cited decisions made at significant cost, such as rejecting government contracts involving domestic surveillance and autonomous weapons development. Nonetheless, AI companies face strong incentives to progress rapidly, especially newer firms like Anthropic and OpenAI, which lack the large, established revenue streams that support giants such as Google and Meta.

With Anthropic planning its IPO and OpenAI reportedly preparing its own public offering, the tension between advancing AI technology and managing its risks remains a central issue in the industry’s evolution.